Wallerian Degeneration: A Major Component of Early Axonal Pathology in Multiple Sclerosis

Abstract

Axonal loss is a major component of the pathology of multiple sclerosis (MS) and the morphological basis of permanent clinical disability. It occurs in demyelinating plaques but also in the so‐called normal‐appearing white matter (NAWM). However, the contribution of Wallerian degeneration to axonal pathology is not known. Here, we analyzed the extent of Wallerian degeneration and axonal pathology in periplaque white matter (PPWM) and lesions in early multiple sclerosis biopsy tissue from 63 MS patients. Wallerian degeneration was visualized using an antibody against the neuropeptide Y receptor Y1 (NPY‐Y1R). The number of SMI‐32‐positive axons with non‐phosphorylated neurofilaments was significantly higher in both PPWM and plaques compared to control white matter. APP‐positive, acutely damaged axons were found in significantly higher numbers in plaques compared to PPWM. Strikingly, the number of NPY‐Y1R‐positive axons undergoing Wallerian degeneration was significantly higher in PPWM and plaques than in control WM. NPY‐Y1R‐positive axons in PPWM were strongly correlated to those in the lesions. Our results show that Wallerian degeneration is a major component of axonal pathology in the periplaque white matter in early MS. It may contribute to radiological changes observed in early MS and most likely plays a major role in the development of disability.

INTRODUCTION

Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the central nervous system. Axonal loss, which results in permanent neurological deficits, is an important component of MS pathology.

Evidence for axonal loss within lesions has been shown in several studies. The axonal density in MS plaques is significantly reduced compared with periplaque white matter (PPWM) (1) and normal controls (26). Axonal transections, indicating irreversible axonal damage, are a consistent feature of MS lesions and correlate with the degree of inflammation within the plaques (35). Axonal transport disturbances identified by staining against amyloid precursor protein (APP) show acute axonal damage, which may or may not be reversible. Acute axonal damage occurs in actively demyelinating lesions and, to a lesser degree, in chronic lesions 1, 16, 22. The number of APP‐positive axons correlates with the number of CD8‐positive T cells and cells of the macrophage/microglia lineage 1, 22, 23. Axonal damage in MS seems to be a continuous process, with the highest number of APP‐positive axons observed during the first years of disease (23).

Axonal loss in normal‐appearing white matter (NAWM) of MS patients has not received much attention up to now. Several studies demonstrated axonal loss in NAWM of spinal cord in MS patients 2, 20, 26. In patients with long‐standing MS, the axonal density in NAWM measured in cervical spinal cord did not differ significantly from axonal density in chronic plaques, although the latter was slightly lower (26). It was suggested that the equally decreased axonal density in the NAWM and in plaques could represent the final outcome of multi‐segmental, long‐standing inflammation affecting the tracts by Wallerian degeneration. The reduced axonal density in MS was also demonstrated in NAWM of corpus callosum and anterior visual pathways 13, 15. The number of SMI 32‐positive terminal axonal ovoids was significantly higher in NAWM from MS brains than in controls (17 vs. 0.7 ovoids/mm³) (35). A diffuse axonal injury as shown by neurofilament‐positive spheroids was found in late disease stages (chronic progressive multiple sclerosis) on the background of a global inflammatory response (24). Kornek et al found a low but significant number of APP‐positive axons in PPWM of MS cases with actively demyelinating lesions compared to controls (22). In these cases, axonal injury was observed in normal WM far distant from established plaques. In inactive cases some increase in the number of APP‐positive axons was found in the PPWM, but it did not reach statistical significance. APP immunoreactivity in normal WM of inactive MS cases was comparable to that in control WM (22). Taken together, these results suggest that axonal injury in MS is not only limited to demyelinating lesions, but also affects the so‐called NAWM.

Wallerian degeneration represents the process of anterograde degeneration of the distal part of the axon that is separated from its cell body. Axonal damage, as indicated by neurofilament dephosphorylation and axonal transport disturbances, might result in transection of axons. The result is Wallerian degeneration and, consequently, axonal loss. Axonal loss in the NAWM might be the result of different processes. It may be caused by inflammatory damage in the NAWM (24). The loss of axons in WM may result from Wallerian degeneration of axons that are transected in MS lesions. There are several neuropathological studies indirectly supporting the latter hypothesis. The discontinuous staining of axonal neurofilaments and the presence of terminal axonal ovoids suggest Wallerian degeneration 12, 35. Evangelou et al have examined the relationship between demyelinating lesion load in cerebral WM of MS patients and the loss of axons in NAWM of the corresponding regions in the corpus callosum (14). They found a strong inverse correlation between the regional lesion load and axonal density in the corresponding NAWM. In another study, a demyelinated lesion located at the cervicomedullary junction in a patient with MS of short duration caused significant axonal loss in NAWM distal to the lesion (3). Moreover, results of neuroradiological studies provide evidence of Wallerian degeneration in NAWM in early stages of MS 6, 8, 33.

The aim of our study was to assess the extent of axonal damage and the contribution of Wallerian degeneration to axonal loss in lesions and PPWM by investigating biopsy tissue from patients with MS of short duration. Understanding axonal pathology is essential for understanding clinical disability.

The focus of the present study was to directly visualize and quantify the number of axons undergoing Wallerian degeneration. As a novel tool, we used an antibody against the neuropeptide Y receptor Y1 (NPY‐Y1R). This antibody stains degenerated nerve fibers (31) and has not been investigated in MS tissue before. Our results show widespread Wallerian degeneration in early MS lesions and PPWM. Wallerian degeneration in lesions and PPWM correlate. Thus, it is highly likely that Wallerian degeneration in PPWM is caused by axonal transection occurring within the lesions.

MATERIALS AND METHODS

Patients

We investigated biopsy tissue from 63 patients who had been diagnosed with inflammatory demyelination of the central nervous system (CNS) consistent with multiple sclerosis. The biopsies had been performed in different neurosurgery centers for various diagnostic reasons, for example, to exclude neoplastic or infectious diseases. Specimens were sent to the Department of Neuropathology in Göttingen, Germany for a second opinion. Clinical background data are summarized in Table 1.

Table 1.

Patient characteristics.

Age: median: 35 years, range: 10–72

Sex: female: 65.1%, male: 34.9%

Clinical diagnosis:

Clinically isolated syndrome suggestive of MS: 38

Relapsing‐remitting: 19

Secondary progressive: 6

Time from first symptoms to biopsy: median: 1.9 months, range: 3.6 days–19 years

The control group consisted of four patients who underwent surgery for temporal lobe epilepsy (median age: 34.5 years, range: 25–41 years; two women, two men). Neuropathological examination revealed no significant abnormalities except for mild astrogliosis in one case.

To assess the impact of remyelination on axonal damage, autopsy material containing shadow plaques from four patients with clinical and neuropathological diagnosis of multiple sclerosis was analyzed (median age: 44 years, range: 35–63 years; three female, one male; diagnosis: relapsing‐remitting MS—one, primary progressive MS—one, secondary progressive MS—two).

Histopathology

Specimens were fixed in 4% paraformaldehyde and embedded in paraffin. Slices 4 µm thick were stained with haematoxylin and eosin (HE), Luxol‐fast blue/periodic acid–Schiff (LFB/PAS) and Bielschowsky's silver impregnation. Immunohistochemical staining was performed with a biotin avidin or an alkaline phosphatase/anti‐alkaline phosphatase technique. The following primary antibodies were used for diagnostic purposes and staging of lesions: anti‐myelin basic protein (anti‐MBP, Boehringer Mannheim, Mannheim, Germany), anti‐proteolipid protein (anti‐PLP, Biozol, Eching, Germany), anti‐myelin oligodendrocyte glycoprotein (anti‐MOG, clone 8‐18‐C5, kindly provided by Prof. Linington, University of Aberdeen, UK), anti‐myelin‐associated glycoprotein (anti‐MAG, kindly provided by Prof. Schwab, University of Zürich, Switzerland), anti‐cyclic nucleotide phosphodiesterase (anti‐CNPase, Sternberger, MD, USA), KiM1P (macrophages, kindly provided by Prof. Radzun, University of Göttingen, Germany) and anti‐MRP 14 (activated macrophages, BMA Biomedicals, Augst, Switzerland), anti‐CD3 (T cells, Serotec, UK), anti‐CD8 (cytotoxic T cells, Dako, Denmark), anti‐IgG (plasma cells, Dako, Denmark) and anti‐complement C9neo antigen (anti‐C9 neo poly, kindly provided by Prof. Morgan, University of Cardiff, UK). To assess axonal injury the following antibodies were used: anti‐phosphorylated neurofilaments (SMI‐31, Sternberger, MD, USA), anti‐non phosphorylated neurofilaments (SMI‐32, Sternberger, MD, USA), and anti‐amyloid precursor protein (anti‐APP, Millicore, MA, USA). To visualize Wallerian degeneration we used an antibody against the neuropeptide‐Y1 receptor (anti‐NPY‐Y1R, Acris Antibodies, Hiddenhausen, Germany). Control stainings were performed in biopsy material from a patient with Wallerian degeneration caused by subacute ischemic stroke (Figure 4C).

Figure 4.

Wallerian degeneration in PPWM, lesions and control white matter: High numbers of axons undergoing Wallerian degeneration are found in PPWM (A) and lesions (B) of MS patients. As control, high numbers of degenerating axons are found in a patient with cerebral infarction (C) but not in patients who had surgery for epilepsy (D). Original magnification: ×400. Scale bar: 50 m. NPY‐Y1R staining, patient #202, inactive lesion. 210 × 297mm (200 × 200 DPI).

Validation of neuropeptide Y‐Y1 receptor immunohistochemistry

To confirm that the anti‐NPY‐Y1R antibody stains nerve fibers undergoing Wallerian degeneration, we analyzed experimental sciatic nerve transection in mice (Figure 1A,B). In addition, we stained sural nerve biopsy tissue of a patient with chronic inflammatory axonal polyneuropathy (CIAP) showing signs of Wallerian degeneration in semithin sections and teased‐fiber preparations (Figure 1C–E). Finally, results were confirmed in Wallerian degeneration slow mice (WLDs mice) suffering from MOG_35‐55 induced experimental autoimmune encephalomyelitis (EAE) (Figure 1F–H).

Figure 1.

Anti‐NPY‐Y1R antibody detects Wallerian degeneration in human and experimental PNS and CNS tissue: (A,B) Mouse sciatic nerve 6 days after transection immunostained with anti‐NPY‐Y1R antibody shows immunolabeled elongated and ovoid structures (B, brown, arrows) not seen in non‐transected control nerve (A). Human sural nerve biopsy showing single NPY‐Y1R‐positive structures (C, red, arrows). Wallerian degeneration is present in this sural nerve biopsy confirmed by toluidin‐blue stained semi‐thin sections (D, arrow) and teased‐fiber preparation (E). (F–H) NPY‐Y1R immunoreactive profiles in wt (F, arrows) and WLDs mice with EAE at peak of disease (G arrow). Significantly more NPY‐Y1R immunoreactive profiles are detected in wt mice compared to WLDs mice at a similar stage of lesion formation (P < 0.05; H). Original magnifications: A,B, D–G: ×400; C: ×200. Scale bars: A,B, D–G: 50 m, C: 100 m. 209 × 297 mm (300 × 300 DPI).

Classification of multiple sclerosis lesions

All biopsy specimens fulfilled the generally accepted criteria for the pathological diagnosis of multiple sclerosis showing an inflammatory demyelinating lesion 24, 32. Lesions were classified according to their demyelinating activity as described in detail earlier (5). This previous work showed that early active (EA) lesions are infiltrated by numerous macrophages that are immunoreactive with all major myelin proteins (MBP, PLP, MAG, MOG, CNP). In late active (LA) lesion areas, demyelination is more advanced and macrophages contain MBP‐ but not MOG‐positive myelin debris. Inactive lesions (IA) are completely demyelinated, and macrophages no longer contain myelin protein‐positive degradation products within their cytoplasm. In early remyelinating plaques, thin, irregularly formed myelin sheaths are seen, as well as a pronounced infiltration by macrophages/microglial cells and T cells. In late remyelinating lesions, remyelination is more advanced and only few inflammatory cells can be found. The PPWM does not show any signs of demyelination.

We investigated autopsy tissue revealing late remyelinated lesions. Remyelination was advanced with formation of so‐called shadow plaques. Shadow plaques in LFB/PAS staining are characterized by myelin pallor due to abnormally thin myelin sheaths with minor residual inflammation.

Early active demyelinating lesions can be divided into subtypes according to their presumed immunopathogenesis. The immunopathological pattern of lesions was determined as previously described (27). Briefly, pattern I and pattern II lesions show a T cell/macrophage‐associated demyelination. The loss of MAG and other myelin components (MBP, MOG, PLP) is comparable. In pattern II, but not in pattern I, demyelination is associated with immunoglobulin and complement deposition at sites of active myelin destruction. In contrast, pattern III lesions show signs of a distal oligodendrogliopathy. Apoptotic oligodendrocytes are found in periplaque white matter. Demyelinated lesions show a preferential loss of MAG or CNPase.

Morphometry

The relative axonal density of Bielschowky silver‐impregnated axons, as well as axons with SMI‐31‐ and SMI‐32‐positive neurofilaments, was determined by point sampling using a 25‐point Zeiss eyepiece. All measurements were performed at 800‐fold magnification and at least 10 microscopic fields were analyzed, depending on sample and lesion size. Random points were superimposed on the plaques and PPWM. Relative axonal density was expressed as percentage of axons crossing the stereological grid points to total number of grid points. The median value of each patient was used for further statistical analysis, avoiding skewing of data due to differing numbers of analyzed microscopic fields.

The number of APP‐ and NPY‐Y1R‐positive axons was counted in at least 10 standardized microscopic fields of 0.01 mm², each defined by an ocular morphometric grid. Discontinously stained fibers were counted only once. In experimental spinal cord tissue, the density of NPY‐Y1R‐positive axons was determined in at least two EAE lesions in the dorsal column in three wild‐type and three WLDs mice. In the text and figures, the median number of APP‐ and NPY‐Y1R‐positive axons/mm² is given.

Statistics

For statistical analysis, non‐parametric tests were performed (Mann–Whitney test, Spearman rank correlation, Kruskal–Wallis ANOVA). In cases of multiple comparisons, a Bonferroni adjustment was applied. All tests were classified as significant if the P value was <0.05. The calculations were performed using the commercial statistical package STATISTICA for Windows, v.8.0 (StatSoft, Inc., Tulsa, OK, USA).

RESULTS

Anti‐NPY‐Y1R antibodies detect Wallerian degeneration in experimental sciatic nerve transection and human sural nerve biopsies

NPY‐Y1R antibodies have been described to label axons undergoing Wallerian degeneration (31). To validate our antibody, we first stained mouse sciatic nerves 6 days after experimental transection (Figure 1A–B). Within the nerve segments distal to the transection side, elongated and ovoid structures were stained by the anti‐NPY‐Y1R antibody (arrow). Similar staining patterns were observed in human sural nerve biopsies with evidence of abundant Wallerian degeneration in semithin sections (Figure 1C, D) and teased‐fiber preparations (Figure 1E).

Fewer NPY‐Y1R‐positive axonal profiles in mice with delayed Wallerian degeneration during EAE

To further confirm that anti‐NPY‐Y1R specifically labels axons undergoing Wallerian degeneration, we examined spinal cord tissue of Wallerian degeneration slow (WLDs) mice suffering from MOG_35‐55‐induced EAE. Numerous NPY‐Y1R‐positive axonal structures were observed primarily in the lesions of wild type mice at peak of disease, whereas significantly fewer NPY‐Y1R axonal structures were found in WLDs mice in the same lesion stage (Figure 1F–H).

Pronounced axonal damage in lesions and high numbers of non‐phosphorylated neurofilaments in PPWM

Axonal pathology was quantified in 44 lesion areas (33 early active, 11 inactive; 4 biopsies contained early active and inactive lesion areas) and in 40 areas of PPWM from 40 biopsies. As controls, WM of four patients who underwent surgery due to epilepsy was analyzed.

Relative axon density was assessed using Bielschowsky's silver impregnation (2, 3) and was found to be significantly lower in plaques than in PPWM (P < 0.01) and control WM (P < 0.01). Lesions showed a 36.8% reduction in axonal density compared to PPWM and a 41.4% reduction compared to control WM. The axonal density in PPWM appeared to be 7.2% lower than in control white matter; however, no significant difference was found (P = 0.47).

Figure 2.

Axonal damage in plaques, periplaque white matter and controls: (A) Bielschowsky's silver impregnation. Relative axonal density in plaques was significantly lower than in PPWM (P < 0.01) and in control WM (P < 0.01). (B) SMI‐31 staining. Relative axonal density in stainings for phosphorylated neurofilaments was significantly lower in lesions compared to PPWM (P < 0.01) and control WM (P = 0.03). (C) SMI‐32 staining. Higher relative axonal density in stainings for SMI‐32‐positive neurofilaments was shown in both PPWM and plaques compared to control WM (P < 0.01 and P = 0.01, respectively). (D) APP staining. The number of APP‐positive axons was significantly higher in plaques compared to PPWM (P < 0.01) and control WM (P < 0.01). (E) NPY‐Y1R staining. The number of NPY‐Y1R‐positive axons was significantly higher in PPWM (P < 0.01) and plaques (P = 0.01) than in control WM. P values are the results of Kruskal‐Wallis test with three groups. *P < 0.05. Relative axonal density was expressed as percentage of axons crossing the stereological grid points to total number of grid points.

Figure 3.

Axonal density and axonal damage in periplaque white matter and lesions: Bielschowsky's silver impregnation shows a significantly higher axonal density in PPWM (A) compared to lesions (B). The same is shown by staining for phosphorylated neurofilaments, although numbers are in general lower (SMI‐31, C = PPWM, D = lesion). Injured axons with non‐phosphorylated neurofilaments, as demonstrated by staining with SMI‐32‐antibodies, are found in high numbers in PPWM (E) as well as in lesions (F). However, acutely damaged axons are found in higher numbers in lesions (H) compared to PPWM (G) (APP staining). Original magnification: ×400. Scale bar: 50 m. Patient #262, early active lesion. 210 × 297mm (600 × 600 DPI).

The relative density of axons with phosphorylated neurofilaments (“healthy axons”) was investigated using the SMI‐31 antibody (2, 3); only about half the number of neurofilaments compared to the Bielschowsky's silver impregnation was observed. Axonal density in plaques was significantly lower than in PPWM (P < 0.01) and control WM (P = 0.03). There was no significant difference in axonal density between PPWM and control WM (P = 1.0). The relative reduction in axonal density was similar when comparing Bielschowsky's silver impregnation and SMI‐31 staining.

For analyzing non‐phosphorylated neurofilaments (“injured axons”), the SMI‐32 antibody was used (2, 3). The relative density of axons with SMI‐32‐positive neurofilaments was higher in both PPWM and plaques compared to control WM, where almost no non‐phosphorylated neurofilaments were found (P < 0.01 and P = 0.01, respectively). PPWM and plaques showed the same median axonal density values (7.8%, 7.8%) with no significant difference seen (P = 1.0).

Acute axonal damage was determined using APP staining (2, 3). The number of APP‐positive axons was significantly higher in plaques compared to PPWM (P < 0.01) and control WM (P < 0.01). Acute axonal damage was 7.5‐fold higher in lesions compared to PPWM. There was no significant difference between the number of APP‐positive axons between PPWM and control WM (P = 0.68), although numbers were 2.2 times higher in PPWM comparing these two groups.

Because our control tissue was derived from temporal lobe specimens, we compared axonal values (Bielschowsky's silver impregnation, SMI 31, SMI 32 and APP) from temporal vs. non‐temporal biopsy locations and did not find any significant differences in either the plaques or in the PPWM.

There was also no significant difference between actively demyelinating and inactive demyelinated lesions when comparing the relative axonal density of SMI‐31‐positive neurofilaments [median (interquartiles): 22% (16.0–26.4) vs. 19.6% (6.0–25.6), P = 0.25] and SMI‐32‐positive neurofilaments [median (interquartiles): 8% (3.2–13.6) vs. 4% (1.5–17.2), P = 0.48], as well as the number of APP‐positive axons [median (interquartiles): 730 (390–1820) vs. 560 (140–1120)/mm², P = 0.18]. The relative axonal density using Bielschowsky's silver impregnation was significantly higher in EA lesions than in IA lesions [median (interquartiles): 57.6% (51.6–65.6) vs. 48% (37.6–54.8), P = 0.01].

Widespread Wallerian degeneration in plaques and PPWM

Wallerian degeneration was quantified in 31 lesion areas (21 EA plaques, 10 IA plaques, 1 patient with EA and IA lesion areas) from 30 biopsies and in 28 areas of PPWM from 28 biopsies (Figure 4A,B). Results were compared with WM from biopsies of four patients who underwent surgery due to epilepsy (Figure 4D).

The number of NPY‐Y1R‐positive axons was significantly higher in PPWM (P < 0.01) and plaques (P = 0.01) than in control WM. A median (interquartiles) number of 335 (161.5‐595) NPY‐Y1R‐positive axons/mm² was found in PPWM compared to 200 (67–320) axons/mm² in lesions and 7 (0–12) axons/mm² in control white matter. The number of NPY‐Y1R‐positive axons was higher in PPWM than in plaques; however, this difference was not significant (P = 0.08).

The numbers of NPY‐Y1R‐positive axons did not reveal any significant differences between the temporal vs. non‐temporal biopsy location either within plaques or in PPWM. Moreover, no significant difference between EA and IA lesions was found (median: 200 vs. 190/mm², P = 0.29). There was also no significant difference between PPWM surrounding EA and IA lesions in terms of the number of NPY‐Y1R‐positive axons (median: 395 vs. 230, P = 0.13).

The number of NPY‐Y1R‐positive axons in PPWM strongly correlated with the number of NPY‐Y1R‐positive axons in plaques (R = 0.93, P < 0.01). No significant correlation was found between the number of APP‐positive or SMI‐32‐positive axons in plaques and NPY‐Y1R‐positive axons in PPWM. Also, no significant correlation was found between the number of SMI‐32‐positive neurofilaments and NPY‐Y1R‐positive axons in the PPWM. There was a significant correlation between the number of APP‐positive axons and NPY‐Y1R‐positive axons in plaques (R = 0.44, P < 0.05).

Axonal injury does not differ in immunopathological subtypes

Lesions were classified according to their immunopathological pattern of demyelination (27). To assess whether a difference in the extent of axonal injury between different immunopathological patterns exists, we analyzed 13 lesions with pattern I, 12 lesions with pattern II and 8 lesions with pattern III, together with their surrounding PPWM. Wallerian degeneration was assessed in 21 lesions and PPWM (pattern I—8 lesions, pattern II—6 lesions, pattern III—7 lesions). No significant differences in axonal injury and Wallerian degeneration were found between the plaques of patterns I, II and III and their surrounding PPWM (data not shown).

Axonal injury in early and late remyelinated plaques

Remyelination could temporarily render axons more vulnerable to degeneration (34). To investigate whether early remyelination is associated with an increased risk of axonal injury, we compared the degree of axonal damage between remyelinating plaques from biopsy cases with remyelinated shadow plaques from autopsy cases. Remyelinating plaques from biopsy cases represent an early phase of remyelination. In contrast, shadow plaques reflect the late phase of remyelination. Nine biopsy (nine plaques) and four autopsy cases (six plaques) were used for Bielschowsky's silver impregnation, SMI‐31, SMI‐32 and APP staining, and five biopsy (five plaques) and four autopsy cases (six plaques) were analyzed using NPY‐Y1R staining.

No significant differences between early remyelinating plaques and shadow plaques were observed for the relative density of SMI‐31‐positive neurofilaments [median (interquartiles): 20% (12–22) vs. 18% (12–19.2), P = 0.60], SMI‐32‐positive neurofilaments [median (interquartiles): 4% (3–8.8) vs. 0.8% (0.7–6.8), P = 0.15], and Bielschowsky silver‐impregnated axons [median (interquartiles): 51.6% (39.6–54.8) vs. 43.6% (37.6–54.4), P = 0.73]. The number of APP‐positive axons [median (interquartiles): 560 (290–980) vs. 108 (53–150)/mm², P = 0.02] and NPY‐Y1R‐positive axons [median (interquartiles): 220 (160–250) vs. 0 (0–15)/mm², P = 0.005] was significantly higher in early remyelinating plaques than in shadow plaques.

DISCUSSION

Our present study shows that Wallerian degeneration is a major component of axonal pathology in plaques and the surrounding periplaque white matter in early MS. Axonal loss in the PPWM may be caused by at least two different mechanisms: (1) secondary degeneration caused by axonal transections in the lesions that are present from disease onset on, or (2) a direct attack on axons in the normal‐appearing white matter.

Neuroradiological studies demonstrated that axonal injury is already present in early disease stages in NAWM of MS patients. Proton magnetic spectroscopy studies showed a reduced level of N‐acetyl aspartate (NAA), a marker of axonal integrity, in NAWM 11, 18, 19. Similarly, magnetization transfer imaging studies showed low magnetization transfer ratios in cerebral NAWM of MS patients 17, 28. SMI 32 reacts with a non‐phosphorylated epitope in neurofilament H and visualizes neuronal cell bodies, dendrites and some thick axons. Moreover, it was shown to stain axonal transections and demyelinated axons in multiple sclerosis plaques 7, 35. Previous pathological studies demonstrated the presence of SMI‐32‐positive spheroids and APP accumulation, albeit at much reduced levels compared to plaques, in WM of MS patients 22, 25, 35. In our study, the number of SMI‐32‐positive neurofilaments was significantly higher in PPWM than in control WM, and was comparable to the number of stained neurofilaments within plaques. We did not find a significant difference in APP‐positive axons between PPWM and control WM, although the number of positive axons was higher in the former. Interestingly, in murine experimental autoimmune encephalomyelitis the number of SMI‐32‐positive neurofilaments measured 26 and 40 days after immunization was very similar between lesion and perilesional WM (21). Results suggest that dephosphorylation of neurofilaments is an early event in the pathogenesis of axonal damage. Neurofilament dephosphorylation might precede APP accumulation or indicate a different mechanism of axonal damage independent of transport disturbance.

Previous studies used mainly autopsy material from patients with long‐standing or fulminant MS. Our study based on biopsy material provides evidence that Wallerian degeneration and dephosphorylation of neurofilaments in PPWM is present during early phases of MS. Patients included in this study showed a median disease duration of 1.9 months, and were biopsied because of atypical clinical presentation such as tumor‐like appearance of lesions. However, histology showed inflammatory demyelinating lesions typical for multiple sclerosis. Our finding is in agreement with neuroradiological observations suggesting widespread axonal damage at the earliest clinical stage of MS 11, 18.

Wallerian degeneration of axons transected in demyelinating lesions could provide a significant contribution to axonal loss in MS NAWM; however, only limited in vivo evidence of this potential mechanism has been reported thus far 3, 14, 35. The results of radiological studies using magnetic resonance imaging (33), proton magnetic spectroscopy (6) and diffusion tensor imaging (8) suggest that Wallerian degeneration is present in NAWM in the early stages of MS. Using an antibody against NPY‐Y1R, we showed that Wallerian degeneration is widespread in plaques and PPWM of early MS. The NPY‐Y1R polyclonal antibody was previously studied in animal models of sciatic nerve and spinal cord transection as well as parietal cortex thermocoagulation (31). In the present study, we could confirm that anti‐NPY‐Y1R labels Wallerian degeneration in human and experimental mouse peripheral nerve, in human CNS tissue after stroke as well as in EAE. As expected for a marker detecting Wallerian degeneration, significantly fewer axonal profiles were labeled in WLDs mice in early EAE characterized by delayed Wallerian degeneration. Thus, although the antigen actually detected by the anti‐NPY‐Y1R antibody in degenerating axons remains unknown, this antibody produces an immunolabeling of fibers that undergo Wallerian degeneration. The number of NPY‐Y1R‐positive axons in PPWM correlated strongly with the number of NPY‐Y1R‐positive axons in lesions, suggesting that axonal damage in PPWM arises as a secondary result of lesional pathology. This finding is in agreement with magnetic resonance spectroscopy studies showing an association of Wallerian degeneration in NAWM with axonal transection within lesions (30). High numbers of APP‐positive axons within the lesions might indicate immediate inflammatory damage and possibly higher susceptibility of demyelinated axons to injury. However, this cannot be found in the PPWM with low numbers of APP‐labeled axons. Wallerian degeneration in the central nervous system is a relatively slow process lasting for months to years (36). In contrast, APP staining allows identification of acutely damaged axons for up to 30 days (4). This could explain the lack of correlation between the number of APP‐positive axons within lesions and the number of NPY‐Y1R‐positive axons in PPWM. In addition, APP reactivity might in part indicate a transient disturbance of axonal transport that does not necessarily lead to axonal transection and subsequent Wallerian degeneration.

Mechanisms other than Wallerian degeneration resulting from lesional pathology might participate in axonal damage of PPWM. Although a strong correlation between the regional lesion load and axonal density in corresponding NAWM was found in corpus callosum (14), no such relationship could be demonstrated in the corticospinal and sensory tracts (10). Moreover, radiological studies pointed out that axonal injury begins very early in the course of MS and is seen in patients with low demyelinating lesion load and no significant disability 11, 18. Thus the pathogenesis of axonal injury in MS PPWM may encompass both Wallerian degeneration due to lesional pathology as well as primary damage to axons, leading to diffuse axonopathy, independent of inflammatory demyelination. As potential effector mechanisms, diffuse inflammation of the NAWM (24) or anti‐neurofascin antibodies could be involved. The latter were shown in animal models to selectively target neurofascin at the nodes of Ranvier, resulting in deposition of complement and promotion of axonal damage in the absence of demyelination (29).

Lucchinetti et al demonstrated a histopathological heterogeneity of MS lesions and suggested that different immunpathogenetic mechanisms are involved in demyelination (27). The question arises as to whether axonal damage is also caused by varying immunopathogenetic mechanisms and differs between subtypes. In our study we did not find a significant difference in axonal injury comparing lesions and PPWM of different immunopathological patterns. This may suggest that regardless of the pathogenesis of demyelination, axonal injury uniformly takes place secondarily to inflammation and demyelination. However, our findings should be interpreted with caution because of the limited number of cases.

We found significantly higher numbers of APP‐ and NPY‐Y1R‐positive axons in early remyelinating plaques compared to shadow plaques, representing the late phase of remyelination. A previous study has shown that in early MS more axonal injury can be found in remyelinated lesions than in inactive demyelinated lesions (23). In demyelinated lesions, an aggregation of sodium channels precedes remyelination (9). This accumulation of sodium channels might render axons prone to intraaxonal sodium accumulation upon repeated impulse conduction (34), which may result in axonal damage. Our findings might support the concept that remyelination may temporarily render axons particularly vulnerable to degeneration (34). We are aware that more sophisticated investigations which also examine the inflammatory infiltrate must be carried out to support the hypotheses presented here.

In conclusion, our results show that Wallerian degeneration contributes significantly to axonal loss in lesions and in PPWM in early MS stages. Wallerian degeneration in PPWM is most likely caused by lesional pathology. Other pathogenetic mechanisms may come into play in long‐standing MS, in which WM changes are extensive and may in part occur independently of lesional pathology. Understanding the mechanisms of axonal injury is fundamental to understanding and targeting clinical disability.